Log key information positioning method, system and device

Abstract

The application relates to the technical field of neural networks, in particular to a Log key information positioning method, system and device. The method comprises the following steps: determining problem information, determining a target keyword according to the problem information; screening Log entries in a Log file according to the target keyword; performing text decomposition on the Log entries according to a grammar rule to obtain text sentence patterns of the Log entries; determining a context sentence pattern of each text sentence pattern according to a Log context of the Log entries in the Log file, wherein the context sentence pattern is marked with a word weight; determining a sentence pattern score of each context sentence pattern according to the word weight of the context sentence pattern; determining a target sentence pattern from each context sentence pattern according to the sentence pattern score, wherein the target sentence pattern is used for determining Log key information. The embodiment scheme of the application can quickly and accurately position Log key information from a Log file.

Description

Technical Field

[0001] This invention relates to the field of neural network technology, and in particular to a method, system, and device for locating key information in Log. Background Technology

[0002] Log files, also known as log files, are collections of operations and their results recorded chronologically. Using log files, the cause of device malfunctions can be traced, a process known as debugging. During debugging, identifying key information from the log files is a crucial problem to solve. Summary of the Invention

[0003] In view of this, embodiments of the present invention provide a method, system, and device for locating key information in a Log file, which can locate key information in a Log file relatively quickly and accurately.

[0004] In a first aspect, embodiments of the present invention provide a method for locating key information in a Log, including:

[0005] Determine the problem information, and then determine the target keywords based on the problem information;

[0006] Based on the target keywords, filter the Log entries in the Log file;

[0007] The Log entries are decomposed according to grammatical rules to obtain the text sentence structure of the Log entries;

[0008] Based on the Log entry in the Log context of the Log file, determine the contextual sentence structure of each text sentence, wherein the contextual sentence structure is labeled with a lexical weight;

[0009] Based on the lexical weights of the contextual sentence patterns, a sentence pattern score is determined for each contextual sentence pattern.

[0010] Based on the sentence pattern score, a target sentence pattern is determined from the sentence patterns in each context, and the target sentence pattern is used to determine the key information of Log.

[0011] Optionally, determining the target keywords based on the question information includes:

[0012] Determine the initial keywords based on the question information;

[0013] The initial keywords are expanded using synonyms to obtain expanded vocabulary;

[0014] The target keywords are determined based on the initial keywords and the expanded vocabulary.

[0015] Optionally, the step of decomposing the Log entry according to grammatical rules to obtain the text sentence structure of the Log entry includes:

[0016] The Log entries are segmented according to grammatical rules to obtain the segmented vocabulary of the Log entries;

[0017] Encode the segmented vocabulary of Log entries to obtain word vectors;

[0018] The semantic combination of the word vectors yields the text sentence of the Log entry.

[0019] Optionally, the step of decomposing the Log entry according to grammatical rules to obtain the text sentence structure of the Log entry includes:

[0020] The Log entry is input into a syntax structure decomposition model, which is used to segment the Log entry according to syntax rules to obtain the segmented vocabulary of the Log entry; to encode the segmented vocabulary of the Log entry to obtain word vectors; and to semantically combine the word vectors to obtain the text sentence of the Log entry.

[0021] Optionally, after performing text decomposition on the Log entry according to grammatical rules to obtain the text sentence structure of the Log entry, the method further includes:

[0022] Using a pre-defined semantic learning library, the text sentence structure of the Log entry is ambiguously understood to obtain the expanded sentence structure of the text;

[0023] The extended sentence structure is used to expand the text sentence structure corresponding to the Log entry.

[0024] Optionally, based on the Log entry in the Log context of the Log file, the contextual sentence structure of each text sentence is determined, wherein the contextual sentence structure is labeled with lexical weights, including:

[0025] The Log context and corresponding text sentence of the Log entry are input into the context prediction model. The context prediction model is used to predict the context sentence of the corresponding text sentence based on the Log context, and to label the context sentence with word weights.

[0026] Optionally, determining the sentence structure score for each contextual sentence structure based on its lexical weight includes:

[0027] The contextual sentences labeled with lexical weights are input into the multilayer perceptron (MLP) classification and scoring model, which is used to score the corresponding contextual sentences based on their lexical weights.

[0028] Secondly, embodiments of the present invention provide a Log key information location system, including:

[0029] The semantic matching module is used to determine question information, determine target keywords based on the question information, and filter Log entries in the Log file based on the target keywords.

[0030] The semantic understanding module is used to decompose the Log entries according to grammatical rules to obtain the text sentence structure of the Log entries;

[0031] The text generation module is used to determine the contextual sentence structure of each text sentence structure based on the Log context of the Log file, wherein the contextual sentence structure is labeled with a word weight;

[0032] A classification module is used to determine the sentence structure score of each context sentence structure based on the lexical weight of the context sentence structure; and to determine the target sentence structure from each context sentence structure based on the sentence structure score, wherein the target sentence structure is used to determine Log key information.

[0033] Thirdly, embodiments of the present invention provide a terminal device, including: at least one processor; and at least one memory communicatively connected to the processor, wherein: the memory stores program instructions executable by the processor, and the processor invokes the program instructions to perform the method as described in the first aspect or any one of the first aspects.

[0034] Fourthly, embodiments of the present invention provide a computer-readable storage medium comprising a stored program, wherein, when the program is executed, it controls the device on which the computer-readable storage medium is located to perform the method described in the first aspect or any one thereof.

[0035] In this embodiment of the invention, target keywords can be obtained based on the problem information. These target keywords can then be used to filter log entries in the log file. The filtered log entries can be used to locate the approximate range of key log information. Subsequently, based on the log context, context association is performed on the text sentence patterns determined from the log entries to obtain the most likely contextual sentence patterns for each text sentence pattern in the log file. Then, the key log information can be located based on the sentence pattern score of the contextual sentence patterns. Through the method of this embodiment, information filtering and analysis of log files can be automatically performed, enabling relatively fast and accurate location of key log information. Attached Figure Description

[0036] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0037] Figure 1 A flowchart illustrating a method for locating key log information provided in an embodiment of the present invention;

[0038] Figure 2 This is a schematic diagram of a Log key information location system provided in an embodiment of the present invention;

[0039] Figure 3 A schematic diagram of another Log key information location system provided in an embodiment of the present invention;

[0040] Figure 4 A schematic diagram of a syntax structure decomposition model provided in an embodiment of the present invention;

[0041] Figure 5 A schematic diagram of a context prediction model provided in an embodiment of the present invention;

[0042] Figure 6 This is a schematic diagram of the structure of a terminal device provided in an embodiment of the present invention. Detailed Implementation

[0043] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0044] During debugging based on log files, it is necessary to locate key log information. This key log information can be used to analyze the cause of the current malfunction in the electronic device. To quickly locate key log information from log files, this invention provides a method for locating key log information. This method first filters log entries from the log file, and then determines valid key information based on the filtered log entries. The filtered log entries represent the approximate range of useful information. The valid key information determined based on the log entries can be used to locate the cause of the malfunction. In determining valid key information based on the filtered log entries, a contextual association approach is employed. Contextual association, as the name suggests, involves combining the log's context to arrive at an accurate understanding of the log entry, thereby facilitating the location of key log information. The following will describe the method for locating key log information in this invention in detail with specific embodiments.

[0045] See Figure 1 This is a flowchart illustrating a method for locating key log information according to an embodiment of the present invention. Figure 1 As shown, the processing steps of this method include:

[0046] 101. Determine the problem information and identify target keywords based on it. Optionally, determining target keywords based on the problem information includes: determining initial keywords based on the problem information; expanding the initial keywords with synonyms to obtain extended vocabulary. The target keywords for step 101 can be determined based on the initial keywords and extended vocabulary. For example, if the current problem with an electronic device is that the SIM card is not registered on the network, the initial keywords determined based on this problem information could include: SIM, Service, noService. Optionally, synonyms for SIM could include, for example, UICC, SIM CARD, SIM STATUS, etc. Optionally, synonyms for Service could include, for example, Network, Network Service, Cellular Service, etc. The keywords determined based on the initial keywords and extended vocabulary could include: SIM, Service, noService, UICC, SIM CARD, SIM STATUS, Network, Network Service, Cellular Service, etc.

[0047] In some embodiments, a synonym matching model can be designed. This model can semantically expand the initial keywords to obtain expanded vocabulary. Optionally, the synonym matching model can be implemented as a Word2vec model. The Word2vec model can then semantically expand the initial keywords.

[0048] 102. Filter Log entries in the Log file based on target keywords. Optionally, multiple Log entries can be filtered. Optionally, the filtered Log entries represent a general range of Log key information that can be further filtered.

[0049] 103. Decompose the Log entries according to grammatical rules to obtain the text sentence structure of the Log entries. Optionally, the text sentence structure of the Log entries is the decomposition of the Log entries from a textual semantic level. In some embodiments, the Log entries can be segmented according to grammatical rules to obtain the segmented vocabulary of the Log entries. Then, the segmented vocabulary of the Log entries can be encoded to obtain word vectors, and semantic combination of the word vectors can obtain the text sentence structure of the Log entries.

[0050] In some embodiments, segmenting Log entries according to grammatical rules includes: identifying the subject + predicate + object in the Log entry, and segmenting the Log entry according to the subject + predicate + object. Optionally, if the Log entry is a sentence that does not distinguish between subject + predicate + object, the Log entry can be segmented according to other inherent rules. For example, the Log entry can be segmented according to punctuation marks and / or special symbols contained in the Log entry.

[0051] In some embodiments, Log entries can be input into a syntax decomposition model. The syntax decomposition model can segment the Log entries according to grammatical rules to obtain the segmented vocabulary of the Log entries. The syntax decomposition model can encode the segmented vocabulary of the Log entries to obtain word vectors. Then, the syntax decomposition model can semantically combine the word vectors of the Log entries to obtain the text sentence structure of the Log entries.

[0052] In some embodiments, after obtaining the Log entry, it can be determined whether the Log entry is a complex sentence based on its sentence length and / or whether it contains preset special symbols. If the sentence length of the Log entry exceeds a preset value or the Log entry contains preset special symbols, the Log entry is determined to be a complex sentence; otherwise, the Log entry is determined to be a simple sentence. Optionally, if the Log entry is a complex sentence, it can be input into a syntax decomposition model (SDM) for word segmentation. Optionally, if the Log entry is a simple sentence, it can be segmented according to grammatical rules to obtain segmented vocabulary. The segmented vocabulary of the Log entry is then input into the SDM. The SDM encodes the segmented vocabulary of the input Log entry to obtain word vectors. The SDM then semantically combines the word vectors of the Log entry to obtain the text sentence structure of the Log entry.

[0053] In some embodiments, after decomposing the Log entries according to grammatical rules to obtain the textual sentence structure of the Log entries, a preset semantic learning library can be used to perform ambiguity understanding on the textual sentence structure of the Log entries. Ambiguity understanding refers to different interpretations of the textual sentence structure. Through ambiguity understanding of the textual sentence structure of the Log entries, extended sentence structures can be obtained. These extended sentence structures can be used to expand and add to the textual sentence structure corresponding to the Log entries.

[0054] 104. Based on the Log context of the Log entry in the Log file, determine the contextual sentence structure of each text sentence, wherein the contextual sentence structure is labeled with a lexical weight.

[0055] In this step, each Log entry corresponds to a Log context in the Log file. This step can filter the Log context of a Log entry from the Log file. Alternatively, after determining the Log entries according to step 102, the Log context of each Log entry can be filtered from the Log file. In this step, context association can be performed on the corresponding text sentences based on the Log context of the Log entry to determine the most likely contextual sentence for each text sentence in the Log file. Furthermore, the lexical weight of each word contained in the contextual sentence can be labeled based on the Log context.

[0056] Optionally, embodiments of the present invention may design a context prediction model. After determining the Log context of a Log entry and the corresponding text sentence, the Log context and the corresponding text sentence can be input into the context prediction model. The context prediction model is used to predict the contextual sentence of the corresponding text sentence based on the Log context, and to assign lexical weights to the contextual sentence.

[0057] 105. Determine the sentence score for each context sentence pattern based on its lexical weights. Optionally, the context sentences labeled with lexical weights can be input into a Multilayer Perceptron (MLP) classification and scoring model, which is used to score the corresponding context sentences based on their lexical weights.

[0058] Specifically, the context prediction model may include an encoding unit. The encoding unit is implemented as a bidirectional LSTM model. Each word of the log context is input into the bidirectional LSTM model, which encodes each word of the log context to obtain a context encoding vector. Optionally, the context prediction model includes a context association unit, which is also implemented as an LSTM model. Specifically, this LSTM model supports matrix dot product and softmax algorithms. The text sentence structure output by the grammatical structure decomposition model is used as a key vector and input into the context association unit. The context encoding vector output by the encoding unit is input into the context association unit. The context association unit performs matrix dot product and softmax algorithms on the context encoding vector and the text sentence structure to generate an attention map. The contextual sentence structure of the text sentence structure can be obtained from this attention map, and the contextual sentence structure is labeled with word weights. Optionally, the context prediction model includes a decoding unit. The decoding unit is implemented as a bidirectional LSTM model and is used to decode the output contextual sentence structure labeled with word weights.

[0059] 106. Based on the sentence structure score, determine the target sentence structure from the sentence structures in each context. The target sentence structure is used to determine key information in the Log. Optionally, the vocabulary contained in the target sentence structure can be used to represent key information in the Log. Alternatively, the log entry corresponding to the target sentence structure in the Log file can be used to represent key information in the Log.

[0060] In this embodiment of the invention, initial keywords are obtained from the question information, and expanded vocabulary is obtained by expanding the initial keywords with synonyms. Target keywords for filtering Log entries are obtained based on the initial keywords and expanded vocabulary. Log entries are then filtered using the target keywords. The filtered Log entries can be used to locate the approximate range of key Log information. Subsequently, context association is performed on the text sentence patterns determined based on the Log entries according to the Log context to obtain the most likely contextual sentence patterns for each text sentence pattern in the Log file. Then, the key Log information can be located based on the sentence pattern score of the contextual sentence patterns. Through the method of this embodiment of the invention, information filtering and analysis of Log files can be performed automatically, and key Log information can be located relatively quickly and accurately.

[0061] See Figure 2 This is a schematic diagram of a Log key information localization system provided in an embodiment of the present invention. The Log key information localization system of this embodiment is used to execute the above-described Log key information localization method. Figure 2As shown, the system includes: a word meaning matching module 201, a semantic understanding module 202, a text generation module 203, and a classification module 204. Specifically: the word meaning matching module 201 is used to determine question information, determine target keywords based on the question information, and filter Log entries in the Log file based on the target keywords. The semantic understanding module 202 is used to decompose the Log entries according to grammatical rules to obtain the text sentence patterns of the Log entries. The text generation module 203 is used to determine the contextual sentence pattern of each text sentence pattern based on the Log context of the Log entry in the Log file, and the contextual sentence pattern is labeled with a lexical weight. The classification module 204 is used to determine the sentence pattern score of each contextual sentence pattern based on the lexical weight of the contextual sentence pattern; and to determine the target sentence pattern from each contextual sentence pattern based on the sentence pattern score, the target sentence pattern being used to determine key Log information.

[0062] See Figure 3 This is a schematic diagram of another Log key information location system provided in an embodiment of the present invention. Figure 3 As shown, the semantic matching module 201 includes a keyword extraction unit, a synonym expansion unit, and a filtering unit. The keyword extraction unit determines initial keywords based on the question information. The synonym expansion unit expands the initial keywords with synonyms to obtain expanded vocabulary. Optionally, the synonym expansion unit can integrate a Word2vec model. The synonym expansion unit can perform semantic word expansion on the initial keywords using the Word2vec model. The filtering unit determines target keywords based on the initial keywords and the expanded vocabulary, and filters Log entries from the Log file based on the target keywords.

[0063] like Figure 3 As shown, the semantic understanding module 202 includes a word segmentation unit, a grammatical structure decomposition unit, and an ambiguity correction unit. The word segmentation unit segments the selected Log entries according to grammatical rules to obtain the segmented vocabulary of the Log entries. The grammatical structure decomposition unit encodes the segmented vocabulary of the Log entries to obtain word vectors, and performs semantic combination based on the word vectors to obtain the text sentence structure of the Log entries. Optionally, the grammatical structure decomposition unit can be implemented as a grammatical structure decomposition model. This grammatical structure decomposition model encodes the segmented vocabulary of the Log entries to obtain word vectors, and performs semantic combination based on the word vectors to obtain the text sentence structure of the Log entries. Optionally, this grammatical structure decomposition model can also integrate the function of the word segmentation unit to segment the selected Log entries according to grammatical rules. See also... Figure 4 This is a schematic diagram of a syntax structure decomposition model provided in an embodiment of the present invention. Figure 4As shown, the grammatical structure decomposition model is an LSTM model. Figure 4 As shown, the words contained in the Log entries are input into the LSTM model. The LSTM model encodes each input word into a word vector. The LSTM model uses a skip approach to semantically combine the word vectors. For example... Figure 4 As shown, CNNs are used to output text sentences composed of word vectors. Optionally, CNNs can be simple networks consisting of convolutional layers, pooling layers, and fully connected layers. Alternatively, CNNs can also be implemented as residual networks. Figure 4 As shown, suppose a Log entry is "Carrierconfig changed Try to bind network service". Each word of this Log entry can be input into an LSTM model. The LSTM model can perform word segmentation and recognition, such as identifying the subject, predicate, or object. The LSTM model can then encode and combine the words, finally outputting two text sentences for the Log entry. Text sentence 1 is "Carrierconfig has changed"; text sentence 2 is "Carrierconfig Trys to bind network service".

[0064] Optionally, an ambiguity correction unit is used to perform ambiguity understanding on multiple text sentences output by the grammatical structure decomposition unit using a preset semantic learning library, to obtain extended sentences of the text sentences. The extended sentences are used to expand and add text sentences corresponding to the Log entries.

[0065] In some embodiments, in addition to filtering Log entries based on keywords, the filtering unit also filters the Log context of Log entries from the Log file. For example... Figure 3 As shown, the Log context input text generation module 203 is filtered by the filtering unit.

[0066] like Figure 3 As shown, the text generation module 203 includes an encoding unit, a context association unit, and a decoding unit. The encoding unit encodes each word in the Log context to obtain a context encoding vector. The context association unit predicts the contextual sentence structure of each text sentence based on the context encoding vector of the Log context and determines the word weights of the words in each contextual sentence structure. The decoding unit decodes and outputs the contextual sentence structure annotated with word weights. Optionally, the text generation module 203 can integrate a context prediction model. This context prediction model is used to implement the functions of the encoding unit, context association unit, and decoding unit. See [link to documentation]. Figure 5 This is a schematic diagram of a context prediction model provided in an embodiment of the present invention. Figure 5 As shown, the context prediction model includes an encoding unit. The encoding unit is specifically implemented as a bidirectional LSTM model. Each word of the Log context is input into the bidirectional LSTM model, which encodes each word of the Log context to obtain a context encoding vector. For example... Figure 5 As shown, the context prediction model uses a context association unit, which is specifically implemented as an LSTM model. Specifically, this LSTM model supports matrix dot product and softmax algorithms. Figure 5 As shown, the text sentence structure output by the grammatical structure decomposition model is input as a key vector into the LSTM model. The context encoding vector output by the encoding unit is also input into the LSTM model. The LSTM model performs matrix dot product and softmax algorithms on the context encoding vector and the text sentence structure to generate an attention map. This attention map reveals the contextual sentence structure of the text sentence structure, and the contextual sentence structure is labeled with lexical weights. Figure 5 As shown, the context prediction model includes a decoding unit. The decoding unit is specifically implemented as a bidirectional LSTM model, used to decode the contextual sentences labeled with word weights.

[0067] like Figure 3 As shown, the text generation module 203 is also connected to the ambiguity correction module 205. The ambiguity correction module 205 can use a preset semantic learning library to understand the ambiguity of various contextual sentence patterns, so as to expand and increase the contextual sentence patterns.

[0068] like Figure 3 As shown, the classification module 204 is used to determine the sentence structure score of each context sentence structure based on its lexical weight; and to determine a target sentence structure from each context sentence structure based on the sentence structure score, wherein the target sentence structure is used to represent key information of Log. Optionally, the classification module 204 can integrate an MLP classification and scoring model. The MLP classification and scoring model can be used to score the corresponding context sentence structures based on their lexical weights.

[0069] The Log key information location system of this invention can be implemented using the methods described in the embodiments above. For parts not described in detail in this embodiment, please refer to the relevant descriptions in the method embodiments. The execution process and technical effects of this technical solution are described in the method embodiments, and will not be repeated here.

[0070] It should be understood that Figure 2 and Figure 3The division of the modules in the Log key information location system shown is merely a logical functional division. In actual implementation, they can be fully or partially integrated into a single physical entity, or they can be physically separated. These modules can be implemented entirely in software via processing element calls; they can be fully implemented in hardware; or some modules can be implemented in software via processing element calls, while others are implemented in hardware. For example, modules 201-204 can be separate processing elements, or they can be integrated into a chip in an electronic device. The implementation of other modules is similar. Furthermore, these modules can be fully or partially integrated together, or they can be implemented independently. During implementation, each step of the above method or each of the above modules can be completed through integrated logic circuits in the hardware of the processor element or through software instructions.

[0071] For example, these modules can be one or more integrated circuits configured to implement the above methods, such as one or more Application Specific Integrated Circuits (ASICs), one or more Digital Signal Processors (DSPs), or one or more Field Programmable Gate Arrays (FPGAs). Alternatively, these modules can be integrated together as a System-On-a-Chip (SOC).

[0072] See Figure 6 This is a schematic diagram of the structure of a terminal device provided in an embodiment of the present invention. Figure 6 As shown, the terminal device is presented in the form of a general-purpose computing device. The components of the terminal device may include, but are not limited to: one or more processors 510, a communication interface 520, a memory 530, and a communication bus 540 connecting different system components (including the memory 530, the communication interface 520, and the processor 510).

[0073] The communication bus 540 represents one or more of several bus architectures, including a memory bus or memory controller, a peripheral bus, a graphics acceleration port, a processor, or a local bus using any of the various bus architectures. Examples of these architectures include, but are not limited to, the Industry Standard Architecture (ISA) bus, the Micro Channel Architecture (MAC) bus, the Enhanced ISA bus, the Video Electronics Standards Association (VESA) local bus, and the Peripheral Component Interconnect (PCI) bus.

[0074] Electronic devices typically include a variety of computer-readable media. These media can be any available media that can be accessed by the electronic device, including volatile and non-volatile media, and removable and non-removable media.

[0075] Memory 530 may include computer system readable media in the form of volatile memory, such as random access memory (RAM) and / or cache memory. The electronic device may further include other removable / non-removable, volatile / non-volatile computer system storage media. Memory 530 may include at least one program product having a set (e.g., at least one) of program modules configured to perform the method steps of embodiments of the present invention.

[0076] A program / utility having a set (at least one) of program modules may be stored in memory 530. Such program modules include—but are not limited to—an operating system, one or more application programs, other program modules, and program data. Each or some combination of these examples may include an implementation of a network environment. The program modules typically perform the functions and / or methods described in the embodiments of this specification.

[0077] The processor 510 executes various functional applications and data processing by running programs stored in the memory 530, such as implementing the Log key information location method provided in the embodiments of the invention.

[0078] In a specific implementation, embodiments of the present invention also provide a computer storage medium, wherein the computer storage medium may store a program, which, when executed, may implement some or all of the steps included in the embodiments provided in this application. The storage medium may be a magnetic disk, optical disk, read-only memory (ROM), or random access memory (RAM), etc.

[0079] In a specific implementation, this embodiment of the invention also provides a chip, including: a processor, which is used to execute computer program instructions stored in a memory, wherein when the computer program instructions are executed by the processor, the chip is triggered to execute the Log key information localization method of this embodiment of the invention.

[0080] In a specific implementation, the present invention also provides a computer program product, which includes executable instructions that, when executed on a computer, cause the computer to perform some or all of the steps in the above method embodiments.

[0081] In this embodiment of the invention, "at least one" refers to one or more, and "more than one" refers to two or more. "And / or" describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent the existence of A alone, the simultaneous existence of A and B, or the existence of B alone. A and B can be singular or plural. The character " / " generally indicates that the preceding and following related objects are in an "or" relationship. "At least one of the following" and similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one of a, b, and c can represent: a, b, c, ab, ac, bc, or abc, where a, b, and c can be single or multiple.

[0082] Those skilled in the art will recognize that the units and algorithm steps described in the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of electronic hardware and software. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0083] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0084] In several embodiments provided by this invention, any function, if implemented as a software functional unit and sold or used as an independent product, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0085] The above description is merely a specific embodiment of the present invention. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this invention should be included within the protection scope of this invention. The protection scope of this invention should be determined by the scope of the claims.

Claims

1. A method for locating key information in a log, characterized in that, include: Determine the problem information, and then determine the target keywords based on the problem information; Based on the target keywords, filter the Log entries in the Log file; The Log entries are decomposed according to grammatical rules to obtain the text sentence structure of the Log entries; Based on the Log entry in the Log context of the Log file, determine the contextual sentence structure of each text sentence, wherein the contextual sentence structure is labeled with a lexical weight; Based on the lexical weights of the contextual sentence patterns, a sentence pattern score is determined for each contextual sentence pattern. Based on the sentence pattern score, a target sentence pattern is determined from the sentence patterns in each context, and the target sentence pattern is used to determine the key information of Log.

2. The method according to claim 1, characterized in that, The step of determining target keywords based on the question information includes: Determine the initial keywords based on the question information; The initial keywords are expanded using synonyms to obtain expanded vocabulary; The target keywords are determined based on the initial keywords and the expanded vocabulary.

3. The method according to claim 1, characterized in that, The step of decomposing the Log entry according to grammatical rules to obtain the text sentence structure of the Log entry includes: The Log entries are segmented according to grammatical rules to obtain the segmented vocabulary of the Log entries; Encode the segmented vocabulary of Log entries to obtain word vectors; The semantic combination of the word vectors yields the text sentence of the Log entry.

4. The method according to claim 3, characterized in that, The step of decomposing the Log entry according to grammatical rules to obtain the text sentence structure of the Log entry includes: The Log entry is input into a syntax structure decomposition model, which is used to segment the Log entry according to syntax rules to obtain the segmented vocabulary of the Log entry; to encode the segmented vocabulary of the Log entry to obtain word vectors; and to semantically combine the word vectors to obtain the text sentence of the Log entry.

5. The method according to claim 1, characterized in that, After performing text decomposition on the Log entry according to grammatical rules to obtain the text sentence structure of the Log entry, the method further includes: Using a pre-defined semantic learning library, the text sentence structure of the Log entry is ambiguously understood to obtain the expanded sentence structure of the text; The extended sentence structure is used to expand the text sentence structure corresponding to the Log entry.

6. The method according to claim 1, characterized in that, The step involves determining the contextual sentence structure of each text sentence based on the Log entry within the Log context of the Log file. The contextual sentence structure is labeled with lexical weights, including: The Log context and corresponding text sentence of the Log entry are input into the context prediction model. The context prediction model is used to predict the context sentence of the corresponding text sentence based on the Log context, and to label the context sentence with word weights.

7. The method according to claim 1, characterized in that, The step of determining the sentence structure score for each contextual sentence structure based on its lexical weight includes: The contextual sentences labeled with lexical weights are input into the multilayer perceptron (MLP) classification and scoring model, which is used to score the corresponding contextual sentences according to their lexical weights.

8. A Log key information location system, characterized in that, include: The semantic matching module is used to determine the question information and, based on the question information, determine the target keywords. And for filtering Log entries in the Log file based on the target keywords; The semantic understanding module is used to decompose the Log entries according to grammatical rules to obtain the text sentence structure of the Log entries; The text generation module is used to determine the contextual sentence structure of each text sentence structure based on the Log context of the Log file, wherein the contextual sentence structure is labeled with a word weight; The classification module is used to determine the sentence structure score of each context sentence structure based on the lexical weight of the context sentence structure. And for determining a target sentence pattern from various contextual sentence patterns based on the sentence pattern score, the target sentence pattern being used to determine Log key information.

9. A terminal device, characterized in that, include: At least one processor; as well as At least one memory communicatively connected to the processor, wherein: The memory stores program instructions that can be executed by the processor, which invokes the program instructions to perform the method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a stored program, wherein, when the program is executed, it controls the device on which the computer-readable storage medium is located to perform the method of any one of claims 1 to 7.

Images